Spatial Comparison of Reliabilities of the EPIC and CERES Models in Simulating Sorghum Yields in the U.S. Great Plains.

Different models often produce different results in crop yield simulations, even with the identical input dataset. In this study, we compared two prominent climate impact models, EPIC and CERES, in simulation of historical NASS county level grain sorghum yields in the southern and central U.S. Great Plains. Overall, the EPIC was about 10% more reliable than CERES under all range of climate conditions (normal or extreme). However both models demonstrated great spatial variability and important differences over the region. The EPIC model was generally superior to CERES in majority of the study area while CERES model showed slight advantages in the central NE and KS region. From the results of this study, we identified location, along with model type, as an important factor contributing to uncertainties in climate impact studies